Network Structure

An EtherCAT network is made up of exactly one master and one or more devices (also called slaves). Understanding the role of each, and how they are connected, is the foundation for everything else in EtherCAT.

The master

The master is the sole controller of the network. There is always exactly one. It:

• Owns the wire — only the master initiates communication. Devices never transmit on their own initiative.
• Drives the cycle — the master sends a new process-data frame on every tick of a fixed clock, commonly every 1 ms or every 250 µs.
• Manages startup — the master discovers every device on the network, configures each one, and walks the whole network through a startup sequence before data exchange begins.
• Detects faults — the master monitors whether each device participated correctly in every cycle and can raise an alarm when something goes wrong.

The master is typically a standard PC or embedded Linux board with a dedicated Ethernet port, running master software that handles all of the above.

Devices (slaves)

Devices are the nodes attached to the network: motor drives, digital and analog I/O modules, encoders, sensors, cameras, and so on. A device:

• Processes the frame on the fly — as the master's frame passes through it, the device reads the bytes addressed to it and writes its own data back into the frame, all in hardware, without interrupting the flow.
• Never initiates — devices are fully passive. They only respond to what the master sends.
• Reports an identity — each device stores a vendor ID and a product code in its configuration memory. The master reads these during startup to identify what kind of device it is dealing with.

The logical ring

Although the physical cabling may look like a simple line of daisy-chained devices, EtherCAT treats the segment as a logical ring. The frame leaves the master, passes through every device in sequence, and the last device sends it back along the return path to the master.

This ring property means:

• The frame's round-trip time is proportional to the number of devices, but each device only adds nanoseconds of forwarding delay — the total remains in the low-microsecond range.
• The master always knows exactly how long the round trip takes for the current segment, which it uses to verify that all devices responded.

Device addressing

Devices are identified by their position in the ring — the first device encountered after the master is position 1, the second is position 2, and so on. This positional address is assigned automatically during startup; you do not configure it manually.

Alongside the positional address, each device also carries a vendor ID and product code stored in its configuration EEPROM. The master uses these two values to match devices on the wire to the software drivers registered for them.

Topology options

EtherCAT is flexible about how devices are physically wired. Three layouts are common:

Line

The simplest topology. Each device has an IN port and an OUT port; you connect them one after another in a chain.

Most EtherCAT deployments use a line. It is easy to wire, easy to trace, and requires no additional hardware.

Tree / Star

A branch is created by inserting a junction device (sometimes called a coupler or switch) at any point in the line. Downstream devices connect to the branch port of the junction.

Tree topologies are useful when devices are spread across physically separate parts of a machine and routing a single long chain would be inconvenient.

Ring (redundancy)

The OUT port of the last device is connected back to a second Ethernet port on the master, closing a physical ring. If a cable breaks, the master can detect the break and switch to the remaining path, keeping the network running.

Ring topology requires master software that supports redundancy and a second NIC on the master, but adds fault tolerance for critical installations.

Next steps

With the network structure in mind, the next topic is How Data Flows — the two kinds of communication that happen over this network and when each is used.